(I typed enter too soon)

I can see a hierarchy of "spam" alerters, good actors who spend time blacklisting bad actors. As this requires decentralised trust you will need a way for the community to "verify" which blacklisters are good and which are bad.

A challenge but not an impossibility...